Automatic analysis of microearthquake network data

Abstract

The microearthquake data collected by the dense network of seismographs in California may contain a wealth of information about the tectonic process, structure and properties of the earth's crust. However, before this data can be fully exploited for the purposes of earthquake prediction research, a relatively automatic, accurate, and uniform procedure for the analysis of seismic data is needed. A simple algorithm which identifies earthquakes and times first arrivals using data from an array of single component seismometers is being studied. The algorithm is simple enough that it could easily be implemented on a microprocessor monitoring a single station in real time. A unique aspect of the algorithm is that it not only picks arrivals but also objectively determines arrival time accuracy. Earthquakes are located by an iterative method. First, the most accurate arrivals are used to locate the event. Then arrivals which do not agree with this location are repicked using a more sensitive algorithm. This process is repeated until a stable location is obtained. For clear, unambiguous arrivals, the arrival times determined by the algorithm agree to within a few hundredths of a second of arrival times determined by an analyst. For weaker, more ambiguous arrivals, computer picks and hand picks are always within the accuracy prescribed by the algorithm. Using such an algorithm in routine processing of seismic data could require as little as 2 percent of the human effort now required by manual processing, and would provide for the first time, an accurate objective data base for earthquake prediction research. Syntactic structural analysis (a formal method of symbolic description) of seismic wave-forms is being investigated as a tool to improve the descriptive capability of the algorithm and as a formalism in which other picking algorithms can be expressed easily and compared. Using syntactic procedures to automatically provide descriptions of the first-arrival waveform is proposed as way of providing useful information about the earth that is not presently available on a routine basis.